Spikes - revisited

Back to the old topic of those spikes that at least some of us see from time to time.
I have a new neighboring station. just a few miles away, with a small “mountain” between us.

I saw a large spike this morning:

This is a Shake-Boom. The spike only occurs on the EHZ channel, no trace of it on HDF (infrasound).

So I thought I would take a look at my neighbor. He has a 3D.
His EHZ channel:

Hmmm… he sees a big spike too, but a lot later than I did. It gets more interesting looking at the other channels:



Not a sign of the spike on those channels.

So it is picked up by the Z channel, but not by N or E channels.
I can only conclude that is is a vertical ground movement, or, just maybe, a very polarized (strong) EMF signal.

Very strong signals, but neither one picked up by the other station, so very local, but strong.

This leads me back to an old theory of mine which is now looking likely (at least to me) again: Both devices are (as far as I am aware) sitting on large concrete floor slabs of out buildings. I think what we are seeing is expansion/contraction due to temperature changes causing stress movements within the concrete (or maybe the buildings themselves?).


Interesting hypothesis! Mine is in the corner of a 36’ishX80ish concrete floored workshop.

1 Like

Mine is approx 30x60 pole barn - concrete floor. Not quite in the corner, but close to the wall.

1 Like

I’ve seen this before on my home made seismometer that was made from accelerometers. It was a 3D accelerometer. After throwing out bad sensor reads, I would get these types of spikes. They would only be for a single sample, and they would only affect one axis at a time. There would be no rebound that you would see from typical motion where you speed up and slow down.

This type of stuff is what you see from ionizing radiation. Most computers harden themselves from this error inducing scenario by using checksums and other fault recovery techniques. What’s typically happening is that a bit is flipped in the sensor read, which can significantly alter the value read from the sensor. The flipped bit is typically overridden by the next sample read. Normally this is a rare occurrence.

If you listen to the Radiolab podcast titled Bit Flip, you’ll hear about other scenarios where computers fail when encountering ionizing radiation.

Ionizing radiation is everywhere. The level of radiation is higher in some places and lower in others. You can shield your sensor to reduce the incidence of it, but it can’t be eliminated. It’s a natural phenomenon that is unrelated to earthquakes.

I did consider cosmic ray hits. It’s a possibility. the reason I disregarded it was because it is not a single sample that shows an extraordinarily high/low value.

Looking at the trace, we see typically something like this:

The recorded data for the period of the spike:

2022-10-26T07:42:30.189999  16704
2022-10-26T07:42:30.199999  16704
2022-10-26T07:42:30.209999  16723
2022-10-26T07:42:30.219999  16674
2022-10-26T07:42:30.229999  16667
2022-10-26T07:42:30.239999  16653
2022-10-26T07:42:30.249999  16217
2022-10-26T07:42:30.259999  15590
2022-10-26T07:42:30.269999  15028
2022-10-26T07:42:30.279999  14476
2022-10-26T07:42:30.289999  14013
2022-10-26T07:42:30.299999  13588
2022-10-26T07:42:30.309999  13157
2022-10-26T07:42:30.319999  12784
2022-10-26T07:42:30.329999  12421
2022-10-26T07:42:30.339999  12094
2022-10-26T07:42:30.349999  11831
2022-10-26T07:42:30.359999  11600
2022-10-26T07:42:30.369999  11344
2022-10-26T07:42:30.379999  11151
2022-10-26T07:42:30.389999  10994
2022-10-26T07:42:30.399999  10775
2022-10-26T07:42:30.409999  10677
2022-10-26T07:42:30.419999  10546
2022-10-26T07:42:30.429999  10452
2022-10-26T07:42:30.439999  10402
2022-10-26T07:42:30.449999  10329
2022-10-26T07:42:30.459999  10308
2022-10-26T07:42:30.469999  10238
2022-10-26T07:42:30.479999  10221
2022-10-26T07:42:30.489999  10165
2022-10-26T07:42:30.499999  10179
2022-10-26T07:42:30.509999  10267
2022-10-26T07:42:30.519999  10257
2022-10-26T07:42:30.529999  10300
2022-10-26T07:42:30.539999  10322
2022-10-26T07:42:30.549999  10384
2022-10-26T07:42:30.559999  10482
2022-10-26T07:42:30.569999  10545
2022-10-26T07:42:30.579999  10609
2022-10-26T07:42:30.589999  10691
2022-10-26T07:42:30.599999  10791
2022-10-26T07:42:30.609999  10846
2022-10-26T07:42:30.619999  10972
2022-10-26T07:42:30.629999  11063
2022-10-26T07:42:30.639999  11144
2022-10-26T07:42:30.649999  11275
2022-10-26T07:42:30.659999  11381
2022-10-26T07:42:30.669999  11449
2022-10-26T07:42:30.679999  11568
2022-10-26T07:42:30.689999  11708
2022-10-26T07:42:30.699999  11830
2022-10-26T07:42:30.709999  11993
2022-10-26T07:42:30.719999  12072
2022-10-26T07:42:30.729999  12218
2022-10-26T07:42:30.739999  12353
2022-10-26T07:42:30.749999  12446
2022-10-26T07:42:30.759999  12624
2022-10-26T07:42:30.769999  12761
2022-10-26T07:42:30.779999  12859
2022-10-26T07:42:30.789999  12965
2022-10-26T07:42:30.799999  13093
2022-10-26T07:42:30.809999  13212
2022-10-26T07:42:30.819999  13351
2022-10-26T07:42:30.829999  13493
2022-10-26T07:42:30.839999  13584
2022-10-26T07:42:30.849999  13683
2022-10-26T07:42:30.859999  13800
2022-10-26T07:42:30.869999  13919
2022-10-26T07:42:30.879999  14082
2022-10-26T07:42:30.889999  14223
2022-10-26T07:42:30.899999  14302
2022-10-26T07:42:30.909999  14435
2022-10-26T07:42:30.919999  14580
2022-10-26T07:42:30.929999  14654
2022-10-26T07:42:30.939999  14762
2022-10-26T07:42:30.949999  14879
2022-10-26T07:42:30.959999  14954
2022-10-26T07:42:30.969999  15088
2022-10-26T07:42:30.979999  15192
2022-10-26T07:42:30.989999  15263
2022-10-26T07:42:30.999999  15398
2022-10-26T07:42:31.009999  15509
2022-10-26T07:42:31.019999  15600
2022-10-26T07:42:31.029999  15727
2022-10-26T07:42:31.039999  15809
2022-10-26T07:42:31.049999  15864
2022-10-26T07:42:31.059999  15996
2022-10-26T07:42:31.069999  16073
2022-10-26T07:42:31.079999  16135
2022-10-26T07:42:31.089999  16251

As you can see the dip (in this case) is not instantaneous, and the recovery even less so.
As far as I know, the counts values are pretty much raw data from the accelerometer. I don’t think there is any sort of filter between which might result in ringing effects. Although … the spikes always have pretty much this form: rapid increase/decrease, with a slower recovery and small overshoot.

Maybe someone with a bit of knowledge of the actual architecture of the digitization board might care to comment on whether there is any sort of filter that might ring (although if so, it is fairly well damped) with sudden step values in readings?

Added: The temperatures have been fairly constant (low to high 50s °F) for the past week, and looking for examples of spikes, I didn’t find any really good examples, just the one above, which is pretty small compared to those seen during the summer when there were much wider temperature swings.

I see interference when operating radio transmitters around my SB (I am a Ham Radio operator). I have suspected that aircraft radio above or near my house could be enough to cause similar anomalies, but have yet to well-correlate this.

1 Like

I see this too - especially when I run about 1Kw :slight_smile:
Not really surprised. I did think about building a faraday screen, with just three holes in the bottom for the R-Shake support “pins” (and of course for the power and Ethernet cables - with ferrite beads inside and out. But at the moment, I don’t do enough operating to make that a priority.

I did look at the possibility of radio-altimeter pulses from aircraft approaching/leaving PDX. But there was just no correlation between overhead aircraft and the spikes - besides, those radio pulses are not that powerful.

A few watts on 160M or 80M is enough for mine to produce spikes. My Shake is in the near field of several antennas.

I’ve tried toroids, and whatever Faraday Cage I have tried to make cannot be made RF-dense enough to stop everything from 160M-70cM without also becoming a resonant drum that causes high incident noise on the shake. I just live with the knowledge that I might miss some things when operating.

My shake and boom did manage to pick up both the seismic event and the sound event from the Tongan eruption this year, so it’s doing well.

1 Like